Department of Ophthalmology, Medical Faculty and University Hospital, University of Cologne, Cologne, Germany.
Division of Dry Eye and Ocular GvHD, University Hospital Cologne, University of Cologne, Cologne, Germany.
Transl Vis Sci Technol. 2022 Jun 1;11(6):24. doi: 10.1167/tvst.11.6.24.
Segmentation and evaluation of in vivo confocal microscopy (IVCM) images requires manual intervention, which is time consuming, laborious, and non-reproducible. The aim of this research was to develop and validate deep learning-based methods that could automatically segment and evaluate corneal nerve fibers (CNFs) and dendritic cells (DCs) in IVCM images, thereby reducing processing time to analyze larger volumes of clinical images.
CNF and DC segmentation models were developed based on U-Net and Mask R-CNN architectures, respectively; 10-fold cross-validation was used to evaluate both models. The CNF model was trained and tested using 1097 and 122 images, and the DC model was trained and tested using 679 and 75 images, respectively, at each fold. The CNF morphology, number of nerves, number of branching points, nerve length, and tortuosity were analyzed; for DCs, number, size, and immature-mature cells were analyzed. Python-based software was written for model training, testing, and automatic morphometric parameters evaluation.
The CNF model achieved on average 86.1% sensitivity and 90.1% specificity, and the DC model achieved on average 89.37% precision, 94.43% recall, and 91.83% F1 score. The interclass correlation coefficient (ICC) between manual annotation and automatic segmentation were 0.85, 0.87, 0.95, and 0.88 for CNF number, length, branching points, and tortuosity, respectively, and the ICC for DC number and size were 0.95 and 0.92, respectively.
Our proposed methods demonstrated reliable consistency between manual annotation and automatic segmentation of CNF and DC with rapid speed. The results showed that these approaches have the potential to be implemented into clinical practice in IVCM images.
The deep learning-based automatic segmentation and quantification algorithm significantly increases the efficiency of evaluating IVCM images, thereby supporting and potentially improving the diagnosis and treatment of ocular surface disease associated with corneal nerves and dendritic cells.
体内共聚焦显微镜（IVCM）图像的分割和评估需要手动干预，既耗时、费力，又不可重复。本研究旨在开发和验证基于深度学习的方法，以自动分割和评估 IVCM 图像中的角膜神经纤维（CNF）和树突状细胞（DC），从而减少分析更大体积临床图像的处理时间。
基于 U-Net 和 Mask R-CNN 架构分别开发了 CNF 和 DC 分割模型；采用 10 倍交叉验证评估两个模型。CNF 模型使用 1097 个和 122 个图像进行训练和测试，DC 模型分别使用 679 个和 75 个图像进行训练和测试，每个折叠 1 次。分析了 CNF 的形态、神经数量、分支点数量、神经长度和扭曲度；对于 DCs，分析了数量、大小和幼稚成熟细胞。编写了基于 Python 的软件，用于模型训练、测试和自动形态计量参数评估。
CNF 模型的平均灵敏度为 86.1%，特异性为 90.1%，DC 模型的平均精度为 89.37%，召回率为 94.43%，F1 得分为 91.83%。手动标注与自动分割之间的组内相关系数（ICC）分别为 0.85、0.87、0.95 和 0.88，用于 CNF 数量、长度、分支点和扭曲度，DC 数量和大小的 ICC 分别为 0.95 和 0.92。
我们提出的方法在 CNF 和 DC 的手动标注与自动分割之间具有可靠的一致性，且速度较快。结果表明，这些方法有可能在 IVCM 图像的临床实践中得到实施。
温迪
